Induced voltage difference processing circuit and muscle fatigue state classification system

Abstract

The utility model relates to an induced voltage difference processing circuit and a muscle fatigue state classification system, and relates to the technical field of induced voltage difference processing circuits and muscle fatigue state classification. The induced voltage difference processing circuit comprises an excitation current circuit, a voltage difference processing circuit and a control circuit, the output end of the excitation current circuit is connected with an electrode array through a multiplexing selection circuit, wherein the electrode array is connected to the outer side of the lower limb of the leg in a sleeving mode through an elastic fabric belt. The electrode array is further connected with an amplification and calculation circuit through the multiplexing selection circuit. Wherein the excitation current circuit is used for providing current excitation for the electrode array; the electrode array is used for collecting induced voltage difference corresponding to internal impedance of tissues of the lower limbs of the legs; and the amplification and calculation circuit is used for amplifying the induced voltage difference and calculating a root mean square corresponding to the amplified induced voltage difference. According to the embodiment of the invention, induced voltage difference processing and muscle fatigue state classification can be realized.

Description

Technical Field

[0001] This disclosure relates to the fields of induced voltage difference processing and muscle fatigue state classification technology, and in particular to an induced voltage difference processing circuit and a muscle fatigue state classification system. Background Technology

[0002] Currently, there is a lack of muscle fatigue state detection devices on the market that possess high accuracy, radiation-free safety, and real-time imaging capabilities. Therefore, this disclosure presents a systematic study on induced voltage difference processing and muscle fatigue state classification, aiming to provide a more reliable and efficient technical means for related fields.

[0003] While traditional methods for detecting muscle fatigue can serve a purpose to some extent, they still have limitations. For example, invasive physiological indicators are often invasive monitoring methods; although the results are accurate, they are costly and cannot achieve real-time monitoring. Furthermore, these methods place a heavy physiological burden on subjects, easily causing discomfort or rejection, limiting their application in daily exercise monitoring and large-scale screening. In addition, currently used non-invasive detection methods also face many challenges in practical applications. For instance, methods for detecting muscle fatigue based on surface electromyography (sEMG) signals: First, in the data preprocessing stage, to extract sufficient information to characterize muscle state, a large number of features are typically extracted from multiple perspectives, including the time domain, frequency domain, and other domains. This leads to high data dimensionality and significant information redundancy, thus prolonging the computation time for subsequent processing and classification, and reducing the real-time performance of the detection. Second, the correlation between some commonly used features is unclear, lacking stable physiological support, which may introduce noise or redundant characteristics in specific application scenarios, thereby affecting the generalization performance and robustness of the classifier. Especially when there are ambiguous boundaries in the state of muscle fatigue, this feature uncertainty is more likely to lead to a decrease in recognition accuracy, which limits the application potential of sEMG method in high-precision fatigue monitoring.

[0004] Electrical Impedance Tomography (EIT) is a traditional medical functional imaging method. Its basic principle involves placing multiple electrodes on the body surface, applying a weak alternating current to some electrodes, and measuring the voltage response on the remaining electrodes. Based on the mathematical relationship between voltage and current, the spatial distribution image of the body's conductivity is obtained. Currently, EIT technology is widely used in various medical fields such as lung function monitoring, brain function imaging, breast cancer screening, and liver detection. Due to its advantages such as fast imaging speed, low cost, non-invasive operation, and no radiation, EIT provides a new technological path for achieving dynamic, non-invasive, and real-time monitoring.

[0005] Calf fatigue, corresponding to the lower limbs, is a common symptom of physical discomfort. Prolonged strenuous exercise such as running or mountain climbing keeps the calf muscles in a contracted state for extended periods, consuming a large amount of energy and producing lactic acid buildup. This stimulates nerve endings, leading to fatigue and weakness in the calves, possibly accompanied by soreness and pain. Additionally, prolonged physical labor or standing / walking for extended periods also causes fatigue due to the continuous tension and contraction of the calf muscles. Therefore, there is an urgent need to develop a circuit that processes the induced voltage difference in the lower limbs to classify calf fatigue. Utility Model Content

[0006] This disclosure presents a technical solution for an induced voltage difference processing circuit and a muscle fatigue state classification system.

[0007] According to one aspect of this disclosure, an induced voltage difference processing circuit is provided, comprising: an excitation current circuit; the output terminal of the excitation current circuit is connected to an electrode array attached to the outer side of the lower limb using an elastic fabric band via a multiplexing selection circuit, the electrode array being further connected to an amplification and calculation circuit via the multiplexing selection circuit; wherein, the excitation current circuit is used to provide current excitation to the electrode array; the electrode array is used to acquire the induced voltage difference corresponding to the internal tissue impedance of the lower limb; the amplification and calculation circuit is used to amplify the induced voltage difference and calculate the root mean square corresponding to the amplified induced voltage difference.

[0008] Preferably, the excitation current circuit includes: a signal generator connected to the electrode array via the multiplexing selection circuit and a voltage-to-current conversion circuit connected to the signal generator; wherein the signal generator is used to generate an excitation voltage supplied to the electrode array; and the voltage-to-current conversion circuit is used to convert the excitation voltage into a corresponding current excitation.

[0009] Preferably, a first filter circuit is provided between the signal generator and the voltage-to-current conversion circuit; wherein the first filter circuit is used to filter the excitation voltage.

[0010] Preferably, the voltage-to-current conversion circuit is configured as a resistive element; wherein one end of the resistive element is connected to the output terminal of the signal generator, and the other end of the resistive element is connected to the electrode array.

[0011] Preferably, the amplification and calculation circuit includes: a programmable differential instrumentation amplifier circuit connected to the electrode array via the multiplexing selection circuit and a root mean square (RMS) detector circuit connected to the programmable differential instrumentation amplifier circuit; wherein, the programmable differential instrumentation amplifier circuit is used to amplify the induced voltage difference; and the RMS detector circuit is used to calculate the root mean square (RMS) of the induced voltage difference amplified by the programmable differential instrumentation amplifier circuit.

[0012] Preferably, a second filter circuit is configured between the programmable differential instrument amplifier circuit and the root mean square detector circuit; wherein, the second filter circuit is used to filter the amplified induced voltage difference.

[0013] Preferably, the induced voltage difference processing circuit further includes: a controller; the output terminal of the controller is connected to the multiplexing circuit, and the multiplexing circuit is also connected to the output terminal of the excitation current circuit, the electrode array disposed on the outer side of the lower limb, the amplification circuit, and the calculation circuit; wherein, the controller is used to control the excitation current circuit to generate current excitation for providing current excitation to the electrode array and for storing or forwarding the root mean square corresponding to the induced voltage difference.

[0014] Preferably, the controller includes: a memory and a processor connected to the memory; wherein the memory is used to store the root mean square corresponding to the induced voltage difference; and the processor is used for the excitation current circuit to provide current excitation to the electrode array and / or to control the multiplexing circuit to perform channel selection switching.

[0015] Preferably, the first electrode in the electrode array is disposed on the lateral side of the tibialis anterior muscle, and the second electrode is disposed on the opposite side of the first electrode; the first group of electrodes and the second group of electrodes remaining in the electrode array excluding the first electrode and the second electrode are respectively evenly distributed on the lateral side of the first leg arc and the second leg arc formed by the first electrode and the second electrode along the lateral side of the leg.

[0016] According to one aspect of this disclosure, a muscle fatigue state classification system is provided, comprising: an induced voltage difference processing circuit as described above; and an electronic device connected to the induced voltage difference processing circuit; wherein the electronic device is used to reconstruct an impedance image of the root mean square corresponding to the induced voltage difference; and a preset classifier connected to the electronic device; wherein the preset classifier is used to classify muscle fatigue state based on the reconstructed impedance image.

[0017] Preferably, the results corresponding to the muscle fatigue state classification include one or more of the following: mild muscle fatigue, moderate muscle fatigue, and severe muscle fatigue.

[0018] Preferably, the preset classifier is configured as a classifier based on a preset convolutional neural network.

[0019] In this disclosure, a technical solution is proposed for an induced voltage difference processing circuit and a muscle fatigue state classification system to solve at least one technical problem in which the prior art cannot detect and process the induced voltage difference of the lower leg corresponding to the leg, thus leading to the inability to classify lower leg fatigue.

[0020] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure.

[0021] Other features and aspects of this disclosure will become clear from the following detailed description of exemplary embodiments with reference to the accompanying drawings. Attached Figure Description

[0022] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the specification, serve to illustrate the technical solutions of this disclosure.

[0023] Figure 1 A schematic diagram of an induced voltage difference processing circuit according to an embodiment of the present disclosure is shown;

[0024] Figure 2 A circuit schematic diagram corresponding to a multiplexing selection circuit according to an embodiment of the present disclosure is shown;

[0025] Figure 3 A circuit schematic diagram corresponding to a sinusoidal AC signal generation circuit according to an embodiment of the present disclosure is shown.

[0026] Figure 4 This is a circuit schematic diagram showing the differential amplifier circuit of the programmable instrument according to an embodiment of the present disclosure;

[0027] Figure 5 A schematic diagram of a muscle fatigue state classification system according to an embodiment of the present disclosure is shown.

[0028] Figure 6 A schematic diagram of the electrode arrangement of the lower limb according to an embodiment of the present disclosure is shown;

[0029] Figure 7 A schematic diagram of a CNN network structure corresponding to a first preset classifier based on a preset convolutional neural network is shown according to an embodiment of the present disclosure. Detailed Implementation

[0030] Various exemplary embodiments, features, and aspects of this disclosure will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0031] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0032] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent three cases: A alone, A and B simultaneously, and B alone. Furthermore, the term "at least one" in this document means any combination of at least two of any one or more elements. For example, including at least one of A, B, and C can mean including any one or more elements selected from the set consisting of A, B, and C.

[0033] Furthermore, to better illustrate this disclosure, numerous specific details are set forth in the following detailed description. Those skilled in the art will understand that this disclosure can be practiced without certain specific details. In some instances, methods, means, components, and circuits well known to those skilled in the art have not been described in detail in order to highlight the main points of this disclosure.

[0034] Figure 1 A schematic diagram of an induced voltage difference processing circuit according to an embodiment of the present disclosure is shown. Figure 1 As shown, the induced voltage difference processing circuit includes: an excitation current circuit 3; the output of the excitation current circuit 3 is connected to an electrode array 1, which is attached to the outside of the lower limb 1-2 of the leg using an elastic fabric strip 1-1, via a multiplexing selection circuit 2; the electrode array 1 is also connected to an amplification and calculation circuit 4 via the multiplexing selection circuit 2; wherein, the excitation current circuit 3 is used to provide current excitation to the electrode array 1; the electrode array 1 is used to collect the induced voltage difference corresponding to the internal tissue impedance of the lower limb 1-2; the amplification and calculation circuit 4 is used to amplify the induced voltage difference and calculate the root mean square of the amplified induced voltage difference. This solves the technical problem that the prior art cannot detect and process the induced voltage difference of the lower leg corresponding to the lower limb.

[0035] In the embodiments of this disclosure and other possible embodiments, the electrode array 1 employs a four-port cyclic excitation mode, sequentially injecting low-amplitude (set amplitude) alternating current into different electrode pairs to obtain the induced voltage difference between unexcited electrode pairs in a cyclic manner. Furthermore, through the multiplexing selection circuit 2 and the amplification and calculation circuit 4, the weak induced voltage difference is accurately captured, ensuring time synchronization and spatial resolution.

[0036] In the embodiments of this disclosure and other possible embodiments, the multiplexing selection circuit 2 includes: four ADG1606 chips; wherein, two ADG1606 chips are dedicated to controlling the excitation channel, and the other two ADG1606 chips are responsible for controlling the measurement channel, thereby realizing a dual-excitation dual-measurement data acquisition mode.

[0037] In the embodiments of this disclosure and other possible embodiments, the multiplexing selection circuit 2 may include: a first multiplexing selection circuit and a second multiplexing selection circuit; the output terminal of the excitation current circuit 3 is connected to the electrode array 1, which is sleeved on the outside of the lower limb 1-2 of the leg using an elastic fabric band 1-1, through the first multiplexing selection circuit, and the electrode array 1 is connected to the amplification and calculation circuit 4 through the second multiplexing selection circuit.

[0038] Figure 2 A circuit schematic diagram corresponding to a multiplexing selection circuit according to an embodiment of the present disclosure is shown. For example... Figure 2 As shown, the multiplexing selection circuit 2, the first multiplexing selection circuit, or the second multiplexing selection circuit includes: multiple multiplexer chips forming the switching channel selection of the electrode array 1. The multiplexing selection circuit 2, the first multiplexing selection circuit, or the second multiplexing selection circuit employs time-division multiplexing technology to divide multiple excitation currents (such as sinusoidal excitation currents) or multiple excitation voltages into time slices and transmit them sequentially, merging them onto a single transmission line to achieve resource sharing and improve bandwidth utilization.

[0039] In embodiments of this disclosure and other possible embodiments, the multiplexer chip may be configured as a multiplexer chip MAX306. For example, the electrode array 1 includes: a first electrode, a second electrode, a third electrode, a fourth electrode, a fifth electrode, a sixth electrode, a seventh electrode, and an eighth electrode.

[0040] In the embodiments of this disclosure, the first electrode in the electrode array 1 is disposed on the lateral side of the tibialis anterior muscle, and the second electrode is disposed on the opposite side of the first electrode; the first group of electrodes and the second group of electrodes remaining in the electrode array 1 excluding the first electrode and the second electrode are respectively evenly distributed on the lateral side of the first leg lower limb arc and the second leg lower limb arc formed by the first electrode and the second electrode along the lateral side of the lower limb 1-2.

[0041] In the embodiments of this disclosure and other possible embodiments, based on the first electrode mark and the second electrode mark, a first leg lower limb arc (first leg lower limb arc segment) and a second leg lower limb arc (second leg lower limb arc segment) are constructed along the lower limb or simulated leg lower limb in clockwise and counterclockwise directions, respectively.

[0042] In the embodiments disclosed herein and other possible embodiments, the first leg lower limb arc and the second leg lower limb arc are connected end to end, and the first leg lower limb arc and the second leg lower limb arc are respectively provided with the first electrode and the second electrode at their ends.

[0043] In the embodiments of this disclosure and other possible embodiments, a first electrode marker corresponding to the tibialis anterior muscle of the leg or simulated leg and a second electrode marker corresponding to the muscle groups of the gastrocnemius, soleus, and tibialis anterior muscles of the leg and lower limb, disposed on the opposite side of the first electrode, are obtained; based on the first electrode marker and the second electrode marker, a first leg and lower limb arc and a second leg and lower limb arc are constructed along the leg or simulated leg and lower limb in clockwise and counterclockwise directions, respectively; based on the number of first electrode groups corresponding to the first leg and lower limb arc and the number of second electrode groups corresponding to the second leg and lower limb arc, a plurality of third electrode markers for the first electrode group corresponding to the number of first electrode groups on the first leg and lower limb arc and a plurality of fourth electrode markers for the second electrode group corresponding to the number of second electrode groups on the second leg and lower limb arc are determined, respectively.

[0044] In embodiments of this disclosure and other possible embodiments, before obtaining the second electrode markers corresponding to the muscle groups of the gastrocnemius, soleus, and tibialis anterior of the leg or simulated leg, which are disposed on the opposite side of the first electrode, determining the second electrode markers corresponding to the muscle groups of the gastrocnemius, soleus, and tibialis anterior of the leg or simulated leg using the first electrode markers corresponding to the tibialis anterior of the leg or simulated leg includes: determining the first electrode markers on the closed horizontal boundary line corresponding to the leg or simulated leg; determining a plurality of candidate electrode markers on the closed horizontal boundary line; calculating a plurality of distances between the first electrode markers and the plurality of candidate electrode markers; configuring the candidate electrode marker corresponding to the largest distance among the plurality of distances as the second electrode markers corresponding to the muscle groups of the gastrocnemius, soleus, and tibialis anterior of the leg or simulated leg, which are disposed on the opposite side of the first electrode.

[0045] In embodiments of this disclosure and other possible embodiments, the step of determining multiple third electrode marks of the first electrode group corresponding to the number of the first electrode group on the first leg lower limb arc and multiple fourth electrode marks of the second electrode group corresponding to the number of the second electrode group on the second leg lower limb arc, based on the number of the first electrode group corresponding to the first leg lower limb arc and the number of the second electrode group corresponding to the second leg lower limb arc, includes: calculating the first length of the first leg lower limb arc with the first electrode mark as the starting point and the second length of the second leg lower limb arc, respectively; determining multiple third electrode marks of the first electrode group corresponding to the number of the first electrode group on the first leg lower limb arc based on the number of the first electrode group corresponding to the first leg lower limb arc and the first length; and determining multiple fourth electrode marks of the second electrode group corresponding to the number of the second electrode group on the second leg lower limb arc based on the number of the second electrode group corresponding to the second leg lower limb arc and the second length.

[0046] In embodiments of this disclosure and other possible embodiments, based on the number of first electrode groups corresponding to the first leg lower limb arc and the first length, determining multiple third electrode marks for the first electrode groups corresponding to the number of first electrode groups on the first leg lower limb arc includes: decomposing the first length into multiple first equal line segments using the number of first electrode groups; and configuring the connection points of adjacent equal line segments in the multiple first equal line segments as third electrode marks at different positions corresponding to each electrode in the first electrode group.

[0047] In embodiments of this disclosure and other possible embodiments, based on the number of second electrode groups corresponding to the second leg lower limb arc and the second length, determining a plurality of fourth electrode marks for the second electrode groups corresponding to the number of second electrode groups on the second leg lower limb arc includes: decomposing the second length into a plurality of second equal line segments using the number of second electrode groups; and configuring fourth electrode marks at different positions corresponding to each electrode in the second electrode group at the connection points of adjacent equal line segments in the plurality of second equal line segments.

[0048] In embodiments of this disclosure and other possible embodiments, the step of constructing a first leg lower limb arc and a second leg lower limb arc along the leg lower limb or simulated leg lower limb in clockwise and counterclockwise directions, respectively, based on the first electrode mark and the second electrode mark, includes: determining a closed horizontal boundary line corresponding to the leg lower limb or simulated leg lower limb parallel to the horizontal plane based on the first electrode mark; constructing a first leg lower limb arc along the closed horizontal boundary line corresponding to the leg lower limb or simulated leg lower limb in clockwise direction, starting from the first electrode mark and ending at the second electrode mark; and constructing a second leg lower limb arc along the closed horizontal boundary line corresponding to the leg lower limb or simulated leg lower limb in counterclockwise direction, starting from the first electrode mark and ending at the second electrode mark.

[0049] In the embodiments of this disclosure and other possible embodiments, the first electrode in the electrode array is disposed at the first electrode mark on the lateral side of the tibialis anterior muscle of the lower leg or simulated lower leg, and the second electrode is disposed at the second electrode mark corresponding to the muscle groups of the gastrocnemius, soleus and tibialis anterior muscles of the lower leg or simulated lower leg disposed on the opposite side of the first electrode; the first group of electrodes and the second group of electrodes remaining in the electrode array excluding the first electrode and the second electrode are respectively evenly distributed on the lateral side of the first lower leg arc and the second lower leg arc formed by the first electrode and the second electrode along the lateral side of the lower leg or simulated lower leg.

[0050] In embodiments of this disclosure and other possible embodiments, the electrode array disposed on the lower limb or simulated lower limb is configured with eight electrodes, including: a first electrode, a second electrode, a third electrode, a fourth electrode, a fifth electrode, a sixth electrode, a seventh electrode, and an eighth electrode. The first electrode is disposed at the first electrode mark A, the second electrode is disposed at the second electrode mark E, and the third, fourth, and fifth electrodes corresponding to the number of the first electrode groups in the electrode array are respectively disposed at multiple third electrode marks B, C, and D corresponding to the arc of the first lower limb; the sixth, seventh, and eighth electrodes corresponding to the number of the second electrode groups in the electrode array are respectively disposed at multiple fourth electrode marks F, G, and H corresponding to the arc of the second lower limb. The first electrode mark A, the second electrode mark E, the multiple third electrode marks B, C, and D, and the multiple fourth electrode marks F, G, and H bisect or equally divide the closed horizontal boundary line formed by the arc of the first lower limb and the arc of the second lower limb.

[0051] In the embodiments of this disclosure and other possible embodiments, the electrode array disposed on the lower limb or simulated lower limb is configured with eight electrodes, including: a first electrode, a second electrode, a third electrode, a fourth electrode, a fifth electrode, a sixth electrode, a seventh electrode, and an eighth electrode. The first, second, third, fourth, fifth, sixth, seventh, and eighth electrodes of the electrode array are arranged in a uniform ring on the lower limb or simulated lower limb using the electrode arrangement method described above for the lower limb. The measurement mode used by the first, second, third, fourth, fifth, sixth, seventh, and eighth electrodes of the electrode array is "adjacent excitation - adjacent measurement," and the excitation current is configured to a set excitation current of 0.01A.

[0052] like Figure 2 As shown, the eight electrodes (first electrode, second electrode, third electrode, fourth electrode, fifth electrode, sixth electrode, seventh electrode, and eighth electrode) can be configured into a switching selection circuit using two 16-to-1 multiplexer chips MAX306. The multiplexer chips include: a first multiplexer chip U4, a second multiplexer chip U8, a third multiplexer chip U5, and a fourth multiplexer chip U9. The 16 output channels of the first multiplexer chip U4, the second multiplexer chip U8, the third multiplexer chip U5, and the fourth multiplexer chip U9 are respectively connected to the first, second, third, fourth, fifth, sixth, seventh, and eighth electrodes of the electrode array 1.

[0053] In an embodiment of this disclosure, the excitation current circuit 3 includes: a signal generator 31 connected to the electrode array 1 via the multiplexing selection circuit 2 and a voltage-to-current conversion circuit 33 connected to the signal generator 31; wherein the signal generator 31 is used to generate an excitation voltage supplied to the electrode array 1; and the voltage-to-current conversion circuit 33 is used to convert the excitation voltage into a corresponding current excitation.

[0054] In embodiments of this disclosure and other possible embodiments, the signal generator 31 may be configured as a direct digital frequency synthesizer corresponding to the design scheme based on the AD9835 chip; the operational amplifier chip of the voltage-current conversion circuit 33 may be configured as an LM358 chip, which has two independent operational amplifiers inside.

[0055] In embodiments of this disclosure and other possible embodiments, the signal generator 31 may also be configured or replaced with a sinusoidal AC signal generating circuit, and the voltage-to-current conversion circuit 33 may be configured or replaced with a current source; wherein the sinusoidal AC signal generating circuit is connected to the current source.

[0056] Figure 3 A circuit diagram corresponding to a sinusoidal AC signal generation circuit according to an embodiment of the present disclosure is shown. Figure 3 As shown, the sinusoidal AC signal generation circuit includes: an accumulator 311, a first adder 312, a second adder 313, a sine lookup table 314, and a digital-to-analog converter 315 connected in sequence; wherein, the accumulator 311 is used to add the frequency control word K and the accumulated frequency control word generated once on the accumulator 311 at each clock edge of the reference clock frequency; the first adder 312 is used to add the accumulated frequency control word and the phase control word P after the number of bits of the added accumulated frequency control word is greater than a set number of bits N; the second adder 313 is used to add the accumulated frequency control word after adding the phase control word P to the waveform control word W; the sine lookup table 314 is used to generate a corresponding sinusoidal discrete signal based on the accumulated frequency control word having the phase control word P and the waveform control word W; the digital-to-analog converter 315 is used to convert the sinusoidal discrete signal into a sinusoidal continuous excitation current.

[0057] In embodiments of this disclosure and other possible embodiments, the sinusoidal AC signal generation circuit further includes: a low-pass filter 316 connected to the digital-to-analog converter 315; wherein, the low-pass filter 316 is used to perform low-pass filtering on the sinusoidal continuous excitation current output by the digital-to-analog converter 315 to obtain the sinusoidal excitation current corresponding to the electrode array 1.

[0058] Figure 4This is a circuit diagram showing the differential amplifier circuit for a programmable instrument according to an embodiment of the present disclosure. For example... Figure 4 As shown, the programmable differential instrumentation amplifier circuit 41 includes: a first-stage programmable differential instrumentation amplifier circuit and a second-stage programmable differential instrumentation amplifier circuit connected in sequence; wherein, the first-stage programmable differential instrumentation amplifier circuit includes: a first-stage chip; the second-stage programmable differential instrumentation amplifier circuit includes: a second-stage chip; the input terminal of the first-stage chip is provided with an RC high-pass filter circuit, and the first-stage chip and the second-stage chip are respectively connected to one end of a corresponding capacitor, the other end of which is connected to analog ground AGND.

[0059] In embodiments of this disclosure and other possible embodiments, the first-stage chip can be configured as an AD8251 manufactured by Analog Devices (ADI), and the second-stage chip can be configured as an AD8253 manufactured by ADI. The AD8251 and AD8253 respectively constitute the first-stage programmable differential instrumentation amplifier circuit and the second-stage programmable differential instrumentation amplifier circuit. An RC high-pass filter circuit is first added before the first-stage programmable differential instrumentation amplifier circuit to filter out the DC bias in the induced voltage difference generated by the electrode array 1. Pin 1 of the first-stage chip is connected to the first voltage output terminal PGAO IN N of the voltage signal generated by the plurality of electrodes 1-1 through the fourteenth capacitor C14. Pin 1 of the first-stage chip is also connected to analog ground AGND through the fifteenth resistor R15. Pin 10 of the first-stage chip is connected to the second voltage output terminal PGAO IN P of the voltage signal generated by the plurality of electrodes 1-1 through the fifteenth capacitor C15. Pin 10 of the first-stage chip is also connected to analog ground AGND through the sixteenth resistor R16. Before pin 3 of the first-stage chip is connected to -5V, it is grounded through parallel capacitors C16 (16th) and C17 (17th). Before pin 8 of the first-stage chip is connected to +5V, it is grounded through parallel capacitors C18 (18th) and C19 (19th). Before pin 3 of the second-stage chip is connected to -5V, it is grounded through parallel capacitors C20 (20th) and C21 (21st). Before pin 8 of the second-stage chip is connected to +5V, it is grounded through parallel capacitors C22 (22nd) and C23 (23rd). Finally, pins 7 of the first-stage chip and pins 7 of the second-stage chip are the programmable differential instrumentation amplification voltages corresponding to the voltage signals generated by the multiple electrodes 1-1.

[0060] In embodiments of this disclosure, a first filter circuit 32 is configured between the signal generator 31 and the voltage-to-current conversion circuit 33; wherein, the first filter circuit 32 is used to filter the excitation voltage.

[0061] In embodiments of this disclosure and other possible embodiments, the voltage-to-current conversion circuit 33 is configured as a resistive element; wherein one end of the resistive element is connected to the output terminal of the signal generator 31, and the other end of the resistive element is connected to the electrode array 1.

[0062] In the embodiments of this disclosure, the amplification and calculation circuit 4 includes: a programmable differential instrumentation amplifier circuit 41 connected to the electrode array 1 via the multiplexing selection circuit 2, and a root mean square (RMS) detector circuit 43 connected to the programmable differential instrumentation amplifier circuit 41; wherein, the programmable differential instrumentation amplifier circuit 41 is used to amplify the induced voltage difference; and the RMS detector circuit is used to calculate the root mean square (RMS) of the induced voltage difference amplified by the programmable differential instrumentation amplifier circuit 41.

[0063] In the embodiments disclosed herein and other possible embodiments, the root-mean-square (RMS) detector circuit 43 is an existing circuit and may include an AD637 chip. That is, the RMS detector circuit is constructed based on the AD637 chip; for details, please refer to https: / / www.vfe.ac.cn / NewsDetail-763.aspx.

[0064] In the embodiments of this disclosure, a second filter circuit 42 is configured between the programmable differential instrumentation amplifier circuit 41 and the root mean square detector circuit 43; wherein, the second filter circuit 42 is used to filter the amplified induced voltage difference.

[0065] In the embodiments of this disclosure, the induced voltage difference processing circuit further includes: a controller 6; the output terminal of the controller 6 is connected to the multiplexing circuit 2, and the multiplexing circuit 2 is also connected to the output terminal of the excitation current circuit 3, the electrode array 1 disposed on the outside of the lower limb 1-2, the amplification circuit, and the calculation circuit 4; wherein, the controller 6 is used to control the excitation current circuit 3 to generate current excitation for providing current excitation to the electrode array 1 and for storing or forwarding the root mean square corresponding to the induced voltage difference.

[0066] In embodiments of this disclosure and other possible embodiments, for example, the controller 6 may be configured as one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), microcontrollers, etc. For example, the controller 6 may be configured as a microcontroller corresponding to an STM32F103ZET6 embedded chip.

[0067] In embodiments of this disclosure, the controller 6 includes: a memory 61 and a processor 62 connected to the memory 61; wherein the memory 61 is used to store the root mean square corresponding to the induced voltage difference; and the processor 62 is used by the excitation current circuit 3 to provide current excitation to the electrode array 1 and / or control the multiplexing circuit 2 to perform channel selection switching.

[0068] In embodiments of this disclosure and other possible embodiments, the memory 61 may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0069] In embodiments of this disclosure and other possible embodiments, a communication component 5 is further configured between the controller 6 or the processor 62 of the controller 6 and the amplification and calculation circuit 4 or the root mean square detector circuit 43 of the amplification and calculation circuit 4.

[0070] In embodiments of this disclosure and other possible embodiments, the communication component 5 is configured to facilitate wired or wireless communication between the controller 6 and other devices. The controller 6 can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, the communication component 5 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, the communication component 5 further includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0071] Figure 5 A schematic diagram of a muscle fatigue state classification system according to an embodiment of the present disclosure is shown. Figure 5 As shown, the muscle fatigue state classification system includes: the induced voltage difference processing circuit as described above; and an electronic device 7 connected to the induced voltage difference processing circuit; wherein the electronic device 7 is used to reconstruct the impedance image of the root mean square corresponding to the induced voltage difference; and a preset classifier 8 connected to the electronic device 7; wherein the preset classifier 8 is used to classify the muscle fatigue state based on the reconstructed impedance image.

[0072] In embodiments of this disclosure and other possible embodiments, the electronic device 7 may be a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness device, personal digital assistant, or other terminal. The electronic device 7 is equipped with an existing impedance image reconstruction module or impedance image reconstruction unit; the impedance image reconstruction module or impedance image reconstruction unit is used to perform impedance image reconstruction on the root mean square corresponding to the induced voltage difference.

[0073] In the embodiments of this disclosure and other possible embodiments, the existing impedance image reconstruction module or impedance image reconstruction unit is configured as an impedance image reconstruction module or impedance image reconstruction unit or impedance image reconstruction system corresponding to the impedance image reconstruction method based on the self-normalized neural network (application number: 202410993963X) or the complex impedance image reconstruction method driven by the space-wavelet dual domain (application number: 2025100080288) or other existing methods or algorithms.

[0074] In embodiments of this disclosure, the results corresponding to the muscle fatigue state classification include one or more of the following: mild muscle fatigue, moderate muscle fatigue, and severe muscle fatigue. Furthermore, the preset classifier 8 is configured as one or more classifiers based on a preset convolutional neural network.

[0075] In the embodiments of this disclosure and other possible embodiments, the classifier based on the preset convolutional neural network can be configured as a classifier corresponding to one or more existing convolutional neural networks such as AlexNet, VGG, GoogLeNet, ResNet, DenseNet, SENet, and ResNeXt.

[0076] In embodiments of this disclosure and other possible embodiments, a muscle fatigue state classification system is characterized by comprising: an excitation current circuit; the output terminal of the excitation current circuit is connected to an electrode array fitted onto the outer side of the lower limb using an elastic fabric band via a multiplexing selection circuit, the electrode array being further connected to an amplification and calculation circuit via the multiplexing selection circuit; wherein, the excitation current circuit is used to provide current excitation to the electrode array; the electrode array is used to acquire the induced voltage difference corresponding to the internal tissue impedance of the lower limb; the amplification and calculation circuit is used to amplify the induced voltage difference and calculate the root mean square corresponding to the amplified induced voltage difference; an electronic device connected to the amplification and calculation circuit; wherein, the electronic device is used to reconstruct an impedance image based on the root mean square corresponding to the induced voltage difference; and a preset classifier connected to the electronic device; wherein, the preset classifier is used to classify muscle fatigue state based on the reconstructed impedance image.

[0077] In embodiments of this disclosure and other possible embodiments, the electronic device is used to perform impedance image reconstruction on the root mean square corresponding to the induced voltage difference, including: solving the root mean square corresponding to the induced voltage difference using an EIT forward problem solver to obtain the measured simulated induced voltage difference or the simulated induced voltage or induced voltage corresponding to the induced voltage difference; and using the electronic device to perform impedance image reconstruction on the simulated induced voltage or induced voltage.

[0078] In the embodiments of this disclosure and other possible embodiments, impedance image reconstruction is performed using the root mean square corresponding to the induced voltage difference of the electrode array to obtain lower limb impedance reconstruction images including the tibialis anterior and gastrocnemius muscles, soleus muscle and tibialis anterior muscle; muscle fatigue state is classified using the lower limb impedance images based on a preset classifier.

[0079] In embodiments of this disclosure and other possible embodiments, the classification of muscle fatigue state based on a preset classifier using the lower limb impedance image includes: a first training unit, used to train the preset convolutional neural network using the lower limb impedance image corresponding to the simulated lower limb and the muscle fatigue state of the lower limb impedance image corresponding to the simulated lower limb, to obtain a first preset classifier corresponding to the preset convolutional neural network; a feature extraction layer, used to extract conductivity distribution features from the lower limb impedance image of the lower limb using the feature extraction layer in the first preset classifier corresponding to the preset convolutional neural network, to obtain the conductivity distribution features of the lower limb; and a first classification unit, used to classify muscle fatigue state based on the classification layer of the first preset classifier, using the conductivity distribution features of the lower limb, to obtain a first muscle fatigue state classification. The first preset classifier can be configured as a classifier corresponding to one or more of the existing convolutional neural networks (CNNs) such as AlexNet, VGG, GoogLeNet, ResNet, DenseNet, SENet, and ResNeXt.

[0080] In embodiments of this disclosure and other possible embodiments, the step of classifying muscle fatigue state based on a preset classifier using the lower limb impedance image further includes: a second classification unit, used to classify muscle fatigue state based on a second preset classifier corresponding to a preset machine learning algorithm, using the root mean square of the induced voltage difference, to obtain a second muscle fatigue state probability value corresponding to the second muscle fatigue state classification; a weighted processing unit, used to weight the first muscle fatigue state probability value and the second muscle fatigue state probability value corresponding to the first muscle fatigue state classification to obtain a muscle fatigue state classification fusion probability distribution vector; and a final muscle fatigue state classification unit, used to determine the final muscle fatigue state classification based on the muscle fatigue state classification fusion probability distribution vector. The second preset classifier corresponding to the preset machine learning algorithm can be configured as one or more of a support vector machine classifier, decision tree classifier, random forest classifier, K-nearest neighbor classifier, logistic regression classifier, adaptive augmentation classifier, linear discriminant analysis classifier, and multilayer perceptron classifier.

[0081] In the embodiments of this disclosure and other possible embodiments, classifying muscle fatigue state using the root mean square corresponding to the induced voltage difference includes: solving the root mean square corresponding to the induced voltage difference using an EIT forward problem solver to obtain the measured simulated induced voltage difference or the simulated induced voltage or induced voltage corresponding to the induced voltage difference; and classifying muscle fatigue state using the simulated induced voltage difference or the simulated induced voltage or induced voltage corresponding to the induced voltage difference.

[0082] In the embodiments of this disclosure and other possible embodiments, before the second preset classifier based on the machine learning algorithm classifies muscle fatigue state using the root mean square corresponding to the induced voltage difference to obtain the second muscle fatigue state probability value corresponding to the second muscle fatigue state classification, a second training unit is used to train the preset machine learning algorithm using the root mean square corresponding to the induced voltage difference of the simulated lower limb to obtain the second preset classifier corresponding to the preset machine learning algorithm; a second classification unit is used to classify muscle fatigue state based on the second preset classifier corresponding to the preset machine learning algorithm using the root mean square corresponding to the induced voltage difference of the lower limb to obtain the second muscle fatigue state probability value corresponding to the second muscle fatigue state classification; a weighted processing unit is used to perform weighted processing on the first muscle fatigue state probability value corresponding to the first muscle fatigue state classification of the lower limb and the second muscle fatigue state probability value of the lower limb to obtain the fusion probability distribution vector of the muscle fatigue state classification of the lower limb.

[0083] In the embodiments of this disclosure and other possible embodiments, the sum of the first muscle fatigue state probability value α and the second muscle fatigue state probability value β of the lower leg is configured to be 1. For example, the first muscle fatigue state probability value α and the second muscle fatigue state probability value β of the lower leg can be configured as 0.8274 and 0.1726, 0.6724 and 0.3276, 0.4291 and 0.5709, 0.2628 and 0.2628, 0.1379 and 0.8621, respectively.

[0084] In the embodiments of this disclosure and other possible embodiments, five rounds of training and testing were conducted on the SVM and CNN models respectively, and the classification accuracy corresponding to different weight values ​​in each round was recorded. By calculating the average performance index of each model in the five rounds, the overall generalization ability was evaluated, and the fusion weights were assigned accordingly. The results corresponding to the first muscle fatigue state probability value α and the second muscle fatigue state probability value β of the lower limb during the five rounds of cross-validation are shown in Table 1.

[0085] The prediction mechanism based on the decision fusion model corresponding to the first preset classifier of the preset convolutional neural network and the second preset classifier of the preset machine learning algorithm is as follows:

[0086] P fusion =αP SVM +βP CNN

[0087] Where α and β represent the first weight value and the second first weight value of SVM (the second preset classifier corresponding to the preset machine learning algorithm) and CNN (the first preset classifier corresponding to the preset convolutional neural network), respectively, and P SVM and P CNN Then, they represent the classification probabilities of each sample in each category by the two models (the first preset classifier corresponding to the preset convolutional neural network and the second preset classifier corresponding to the preset machine learning algorithm) respectively (the first muscle fatigue state probability value corresponding to the first muscle fatigue state classification and the second muscle fatigue state probability value corresponding to the second muscle fatigue state classification), and Pfusion is the muscle fatigue state classification fusion probability distribution vector corresponding to the fused comprehensive score.

[0088] Then, five rounds of training and testing were conducted on the SVM and CNN models respectively, and the classification accuracy corresponding to different weight values ​​in each round was recorded. By calculating the average performance index of each model in the five rounds, the overall generalization ability was evaluated, and the fusion weights were assigned accordingly. The results corresponding to different weight values ​​in the five rounds of cross-validation are shown in Table 1.

[0089] Table 1 shows the classification results corresponding to the probability values ​​α of the first muscle fatigue state and β of the second muscle fatigue state of the lower limb.

[0090] Table 1. Classification results corresponding to different weight values.

[0091]

[0092] Based on the accuracy of the fusion classification, the optimal fusion ratio was determined to be 0.2628 for SVM and 0.7372 for CNN. This indicates that although SVM accounts for a relatively small proportion in this task, it still provides a useful supplement to the final decision of the model, demonstrating its advantages in boundary discrimination and having certain auxiliary prediction value. CNN has stronger capabilities in feature extraction and expression, and has become the main discrimination basis of the fusion model.

[0093] In the classification and recognition process, pre-trained SVM and CNN models are first invoked in parallel to extract features and classify the current input voltage and impedance image data, respectively. Each model independently outputs a corresponding predicted label, representing its classification of the sample. To achieve effective fusion decision-making, these two labels are then uniformly converted into a probability distribution form, where each candidate category corresponds to a probability channel, representing the model's prediction confidence for each category. Subsequently, the probability vectors output by each model are weighted and summed according to preset weight coefficients to obtain a fused weighted probability distribution vector (muscle fatigue state classification fusion probability distribution vector). Finally, the category with the highest fusion probability is selected as the prediction result of the fusion model, achieving weighted integration of multi-model discriminative information.

[0094] Based on the accuracy of the fusion classification, the optimal fusion ratio was determined to be 0.2628 for SVM and 0.7372 for CNN. This indicates that although SVM accounts for a relatively small proportion in this task, it still provides a useful supplement to the final decision of the model, demonstrating its advantages in boundary discrimination and having certain auxiliary prediction value. CNN has stronger capabilities in feature extraction and expression, and has become the main discrimination basis of the fusion model.

[0095] In embodiments of this disclosure and other possible embodiments, before arranging the electrode array on the outer side of the simulated lower limb, the simulated lower limb is constructed, including: a simulated planar structure corresponding to the lower limb; and, within the simulated planar structure, a simulated region of the tibialis anterior muscle and simulated regions of the gastrocnemius, soleus and tibialis anterior muscle groups are constructed to obtain the simulated lower limb.

[0096] In embodiments of this disclosure and other possible embodiments, before classifying muscle fatigue states using the lower limb impedance images based on a preset classifier, the method further includes: simulating different muscle fatigue states of the simulated lower limb using a set lactate conductivity and a set background tissue conductivity, to obtain simulated lower limbs corresponding to different muscle fatigue states.

[0097] Figure 6 This diagram illustrates the electrode arrangement of a leg or lower limb according to an embodiment of the present disclosure. Before acquiring the first electrode marker corresponding to the tibialis anterior muscle of the leg or simulated leg or lower limb, determining the first electrode marker includes: acquiring the tibialis anterior muscle region M1 corresponding to the leg or lower limb or simulated leg or lower limb; configuring any point of the tibialis anterior muscle region M1 as the first electrode marker A corresponding to the tibialis anterior muscle of the leg or lower limb or simulated leg or lower limb; determining the closed horizontal boundary line of the first electrode marker corresponding to the leg or lower limb or simulated leg or lower limb; setting a second candidate electrode marker on the closed horizontal boundary line on the outer side of the leg or lower limb or simulated leg or lower limb furthest from the first electrode marker A; determining the first electrode marker... If the second electrode candidate marker is in the muscle group region M2 corresponding to the gastrocnemius, soleus, and tibialis anterior muscles of the lower leg, then the second electrode candidate marker is configured as the second electrode marker E; otherwise, the first electrode marker A is adjusted according to the set step length and set direction until the second electrode candidate marker is in the muscle group region M2 corresponding to the gastrocnemius, soleus, and tibialis anterior muscles of the lower leg, and the second electrode candidate marker at this time is configured as the second electrode marker E.

[0098] In the embodiments of this disclosure and other possible embodiments, based on the distribution of lactic acid in the muscle compartments corresponding to the simulated tibialis anterior muscle region (the simulated tibialis anterior muscle region M1) and the simulated muscle groups of the gastrocnemius, soleus, and tibialis anterior muscles (the simulated muscle groups of the gastrocnemius, soleus, and tibialis anterior muscles region M2), different muscle fatigue states are divided into three levels: mild muscle fatigue state S1, moderate muscle fatigue state S2, and severe muscle fatigue state S3.

[0099] For example, such as Figure 2As shown, in embodiments of this disclosure and other possible embodiments, the electrode array disposed on the lower limb or simulated lower limb is configured with eight electrodes, including: a first electrode, a second electrode, a third electrode, a fourth electrode, a fifth electrode, a sixth electrode, a seventh electrode, and an eighth electrode. The first electrode is disposed at the first electrode mark A, the second electrode is disposed at the second electrode mark E, and the third, fourth, and fifth electrodes corresponding to the number of the first electrode groups in the electrode array are respectively disposed at multiple third electrode marks B, C, and D corresponding to the arc of the first lower limb; the sixth, seventh, and eighth electrodes corresponding to the number of the second electrode groups in the electrode array are respectively disposed at multiple fourth electrode marks F, G, and H corresponding to the arc of the second lower limb. The first electrode mark A, the second electrode mark E, the multiple third electrode marks B, C, and D, and the multiple fourth electrode marks F, G, and H bisect or equally divide the closed horizontal boundary line formed by the arc of the first lower limb and the arc of the second lower limb.

[0100] In the embodiments of this disclosure and other possible embodiments, the electrode array disposed on the lower limb or simulated lower limb is configured with eight electrodes, including: a first electrode, a second electrode, a third electrode, a fourth electrode, a fifth electrode, a sixth electrode, a seventh electrode, and an eighth electrode. The first, second, third, fourth, fifth, sixth, seventh, and eighth electrodes of the electrode array are arranged in a uniform ring on the lower limb or simulated lower limb using the electrode arrangement method described above for the lower limb. The measurement mode used by the first, second, third, fourth, fifth, sixth, seventh, and eighth electrodes of the electrode array is "adjacent excitation - adjacent measurement," and the excitation current is configured to a set excitation current of 0.01A.

[0101] In the embodiments of this disclosure and other possible embodiments, before simulating different muscle fatigue states of the simulated lower limb using a set lactate conductivity and a set background tissue conductivity to obtain simulated lower limbs (simulated lower limb models) corresponding to different muscle fatigue states, the simulated planar structure corresponding to the cross-section of the lower limb is divided into simulated meshes to obtain multiple simulated meshes corresponding to the simulated planar structure; the set lactate conductivity is set in multiple set continuous simulated meshes in the tibialis anterior muscle simulation region corresponding to the multiple simulated meshes; wherein, the number of multiple set continuous simulated meshes in the tibialis anterior muscle simulation region is greater than 0 and does not exceed the maximum number of set continuous simulated meshes corresponding to the tibialis anterior muscle simulation region; the set background tissue conductivity is set in multiple set continuous simulated meshes in other simulation regions besides the tibialis anterior muscle simulation region to obtain a simulated lower limb corresponding to mild muscle fatigue.

[0102] In the embodiments of this disclosure and other possible embodiments, the set lactate conductivity is set in multiple set continuous simulation grids corresponding to the maximum set number of set continuous simulation grids in the simulated tibialis anterior muscle simulation area within the simulated planar structure; the set lactate conductivity is set in multiple simulation grids corresponding to the muscle group simulation areas of the gastrocnemius, soleus, and tibialis anterior muscles within the simulated planar structure that are smaller than the first set area, thereby obtaining a simulated lower limb corresponding to moderate muscle fatigue.

[0103] In the embodiments of this disclosure and other possible embodiments, the set lactate conductivity is set in multiple set continuous simulation grids corresponding to the maximum set number of set continuous simulation grids in the simulated tibialis anterior muscle simulation area within the simulated planar structure; the set lactate conductivity is set in multiple simulation grids corresponding to the muscle group simulation areas of the gastrocnemius, soleus, and tibialis anterior muscles within the simulated planar structure that are larger than the first set area and smaller than or equal to the second set area, thereby obtaining a simulated lower limb corresponding to severe muscle fatigue.

[0104] In embodiments of this disclosure and other possible embodiments, the step of simulating different muscle fatigue states of the simulated lower limb using a set lactate conductivity and a set background tissue conductivity to obtain simulated lower limbs corresponding to different muscle fatigue states includes: setting the set lactate conductivity in the tibialis anterior muscle simulation area within the simulated planar structure; setting the set background tissue conductivity in other simulation areas besides the tibialis anterior muscle simulation area to obtain a simulated lower limb corresponding to mild muscle fatigue; setting the set lactate conductivity in the tibialis anterior muscle simulation area within the simulated planar structure; and using the gastrocnemius, soleus, and tibialis anterior muscles within the simulated planar structure in an area smaller than a first set area. A set lactate conductivity is set in the muscle group simulation area to obtain a simulated lower limb corresponding to moderate muscle fatigue; the set lactate conductivity is set in the tibialis anterior muscle simulation area within the simulation planar structure; the set lactate conductivity is set in the muscle group simulation areas of the gastrocnemius, soleus, and tibialis anterior muscles within the simulation planar structure that are larger than the first set area and smaller than or equal to the second set area to obtain a simulated lower limb corresponding to severe muscle fatigue; wherein, the first set area is configured as the area corresponding to the tibialis anterior muscle simulation area within the simulation planar structure; the second set area is configured as the sum of the areas corresponding to the muscle group simulation areas of the gastrocnemius, soleus, and tibialis anterior muscles within the simulation planar structure.

[0105] In the embodiments of this disclosure and other possible embodiments, the set lactic acid conductivity and the set background tissue conductivity can be configured to 0.7 S / m and 0.15 S / m, respectively.

[0106] In embodiments of this disclosure and other possible embodiments, determining the final muscle fatigue state classification based on the muscle fatigue state classification fusion probability distribution vector includes: the final muscle fatigue state classification is configured as the muscle fatigue state classification corresponding to the highest fusion probability in the muscle fatigue state classification fusion probability distribution vector.

[0107] In the embodiments of this disclosure and other possible embodiments, in order to construct the EIT simulation forward problem model of the lower leg, this disclosure uses a simulation planar structure corresponding to a two-dimensional circular geometric model to simplify the modeling of the cross-section of the lower leg (leg and lower limb), thereby reducing computational complexity and facilitating the implementation and analysis of subsequent image reconstruction algorithms (using the root mean square corresponding to the induced voltage difference of the electrode array to perform impedance image reconstruction, obtaining the lower limb impedance reconstruction algorithm corresponding to the lower limb impedance reconstruction image including the tibialis anterior muscle, gastrocnemius muscle, soleus muscle, and tibialis anterior muscle). The conductivity of the background tissue (set background tissue conductivity) is set to 0.15 S / m, referencing the average conductivity level of normal muscle tissue. To study the conductivity change characteristics under muscle fatigue, this disclosure constructs lactate distribution models representing three different fatigue levels, and implants lactate regions in the simulation grid corresponding to the simulation planar structure, with their conductivity set (set lactate conductivity) at 0.7 S / m to reflect the influence of lactate accumulation on tissue electrical parameters.

[0108] In the embodiments and other possible embodiments of this disclosure, in the process of reconstructing lower limb impedance images, to improve the stability and spatial resolution of the reconstruction results, the TK-Noser regularization algorithm is selected as the constraint means for solving the model. To reasonably select the regularization parameters and avoid over-smoothing or reconstruction noise amplification, this disclosure uses cross-validation for parameter tuning. Within a preset parameter range, multiple subsets are trained and validated sequentially, and the optimal regularization coefficient is finally determined to be 0.05, which achieves a good balance between reconstruction accuracy and noise robustness.

[0109] In the embodiments of this disclosure and other possible embodiments, in the MATLAB environment, using the EIDORS software package and in conjunction with loop statements, a positive model corresponding to three fatigue states (mild muscle fatigue state S1, moderate muscle fatigue state S2, and severe muscle fatigue state S3) is constructed.The steps for obtaining the simulated induced voltage (simulated induced voltage data) are as follows: (1) Using the above-mentioned electrode arrangement method for the lower limbs, create a simulated planar structure corresponding to the simulated lower limbs with an electrode array (use the mk_common_model function to create an 8-electrode circular model for simulating the human calf); (2) Configure the conductivity corresponding to the simulated planar structure of the simulated lower limbs to the default conductivity setting (use mk_image to create an empty model, at which time the conductivity of the model is the default conductivity 1); (3) According to the lactic acid distribution of the lower limbs, divide the simulated planar structure corresponding to the default conductivity setting into mild muscle fatigue state S1, moderate muscle fatigue state S2 and severe muscle fatigue state S3. The simulated regions of the tibialis anterior muscle, gastrocnemius muscle, soleus muscle, and tibialis anterior muscle corresponding to the maximum lactic acid distribution area in muscle fatigue state S3 are defined. These simulated regions are then stored in the first preset region (polygon_muscle_1), the second preset region (polygon_muscle_2), the third preset region (polygon_muscle_3), and the fourth preset region (polygon_muscle_4), respectively. (This divides the muscle fatigue state into different levels, defining the maximum lactic acid distribution areas and storing them in polygon_muscle_1, polygon_muscle_2, polygon_muscle_3, and polygon_muscle_4). (4) Perform simulation mesh division on the simulation planar structure corresponding to the cross-section of the lower leg, and obtain multiple simulation meshes corresponding to the simulation planar structure. Modify the conductivity of multiple simulated meshes in multiple simulated meshes of the tibialis anterior muscle simulation region, the gastrocnemius muscle simulation region, the soleus muscle simulation region and the tibialis anterior muscle simulation region, respectively, and generate the set group simulation planar structures corresponding to mild muscle fatigue state S1, moderate muscle fatigue state S2 and severe muscle fatigue state S3 respectively (by modifying img.elem_dat in the preset area). (a) Set the lactate conductivity of different types of muscle fatigue states; at the same time, use loop statements to generate 410 sets of data for each type of muscle fatigue state); (5) Use the adjacent excitation adjacent measurement mode to excite and measure the electrode array of the set group simulation plane structure corresponding to mild muscle fatigue state S1, moderate muscle fatigue state S2 and severe muscle fatigue state S3 respectively, solve the EIT positive problem, and obtain the simulation induced voltage corresponding to the measured simulation induced voltage difference (use mk_stim_patterns to set the excitation measurement mode to adjacent excitation adjacent measurement, and then assign it to the set group simulation plane structure corresponding to the three types of muscle fatigue states respectively, and obtain the simulation induced voltage after solving the EIT positive problem).

[0110] Subsequently, image reconstruction was performed using the acquired simulation voltage data in the MATLAB environment to obtain lower limb impedance reconstruction images including the tibialis anterior muscle, gastrocnemius muscle, soleus muscle, and tibialis anterior muscle. The specific steps are as follows: (1) Import the simulation planar structure corresponding to the simulated lower limb with an electrode array, which was created using the above-mentioned electrode arrangement method for the lower limb (import the 8-electrode circular model created using mk_common_model before); (2) Configure the excitation measurement mode to adjacent excitation adjacent measurement mode (use mk_stim_patterns to set the excitation measurement mode to adjacent excitation adjacent measurement); (3) The simulation The conductivity of the simulated planar structure of the lower leg is configured to the default conductivity (a model with the default conductivity is created using mk_image); (4) The regularization algorithm is set and the regularization parameters are determined by cross-validation (the regularization algorithm is set to TK-Noser and the regularization parameters are determined by cross-validation); (5) The simulated induced voltage corresponding to the measured simulated induced voltage difference is solved by image solving to obtain the lower limb impedance reconstruction image including the tibialis anterior muscle and gastrocnemius muscle, soleus muscle and tibialis anterior muscle (after inputting the simulated voltage data, the image is solved by inv_solve and each voltage data is traversed by loop statement).

[0111] Figure 7 A schematic diagram of a CNN network structure corresponding to a first preset classifier based on a preset convolutional neural network is shown according to an embodiment of the present disclosure. Figure 7 As shown, the CNN architecture designed in this disclosure consists of three convolutional modules. Each module includes a convolutional layer, a batch normalization layer, and a ReLU activation function. Max pooling is introduced at the end of the first two modules to progressively compress the feature map size and reduce the parameter scale. To further reduce model complexity and suppress overfitting, a global average pooling layer is used after the third module. The final classification layer consists of a fully connected layer with three neurons, a softmax layer, and a classification output layer to achieve probability prediction and label determination for three types of images.

[0112] The Adam optimization algorithm was employed during model training. This algorithm combines first-order and second-order moment estimation, exhibiting strong convergence efficiency and adaptability. The initial learning rate was set to 0.001 to ensure sufficient weight update amplitude in the early stages of training. A piecewise learning rate descent strategy was introduced, gradually reducing the learning rate during training to improve the model's stability and generalization ability towards the end of convergence. The entire training process consisted of 10 epochs, each using a batch size of 32. Before each iteration, the order of the training samples was randomly shuffled to avoid training bias introduced by data order and improve the model's adaptability to data distribution. The excellent training and validation results further validated the effectiveness of the adopted optimization strategy and training configuration.

[0113] After training, the model achieved a relatively ideal classification accuracy on the validation set, indicating its good generalization ability. In the performance evaluation phase, the model was trained using the MATLAB trainNetwork interface, and the classify function was used to perform classification prediction and performance testing on both the validation set and the actual test dataset. The final classification results show that the model has good discriminative ability.

[0114] In the embodiments of this disclosure and other possible embodiments, an 8-electrode cylindrical water tank is designed as the test object to evaluate the imaging performance of the corresponding induced voltage difference processing circuit of the hardware platform. An agar and a metal piece are placed in the cylindrical water tank to represent low-conductivity and high-conductivity regions in the target field, respectively. Image reconstruction analysis of the experimentally acquired data shows that the reconstruction results can clearly present the spatial distribution characteristics of different conductivity regions in the target field, successfully reflecting the differences between high- and low-conductivity materials in the imaging image. This result not only proves the accuracy and stability of the constructed hardware platform in the voltage data acquisition process but also indirectly verifies the practicality and feasibility of the entire hardware platform in impedance imaging tasks. This provides a solid technical foundation for subsequent data acquisition work based on impedance characteristics in actual muscle fatigue grading detection.

[0115] In the embodiments of this disclosure and other possible embodiments, an electrode array 1, fitted with an elastic fabric strip 1-1 on the outside of the lower limb 1-2, is used for sensing and transmitting voltage signals in the lower limb. An excitation source sequentially acts on different adjacent electrode pairs, forming a specific current path. During this process, the remaining electrode pairs not involved in the excitation are used to collect the induced voltage difference to capture the changing characteristics of the internal impedance distribution of the tissue. Each complete excitation-measurement cycle acquires 40 sets of voltage data, which constitute a valid sample as the basis for subsequent image reconstruction and pattern classification.

[0116] In embodiments of this disclosure and other possible embodiments, it further includes: a terminal device 10, the terminal device 10 being configured with a display screen 10-1; wherein the display screen 10-1 is used to display the muscle fatigue state classification or the final muscle fatigue state classification.

[0117] In the embodiments of this disclosure and other possible embodiments, the terminal device 10 may be configured as a mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness device, personal digital assistant, etc.

[0118] In embodiments of this disclosure and other possible embodiments, display screen 10-1 may include a liquid crystal display (LCD) and a touch panel (TP). If display screen 10-1 includes a touch panel, display screen 10-1 may be implemented as a touchscreen to receive input signals from a user. The touch panel includes one or more touch sensors to sense touches, swipes, and gestures on the touch panel. The touch sensors may sense not only the boundaries of the touch or swipe action but also the duration and pressure associated with the touch or swipe operation.

[0119] In embodiments of this disclosure and other possible embodiments, a wireless communication module 9 is provided between the terminal device 10 and the classifier 8; wherein, the wireless communication module 9 is used to send the muscle fatigue state classification or the final muscle fatigue state classification to the terminal device 10.

[0120] In embodiments of this disclosure and other possible embodiments, the wireless communication module 9 can access wireless networks based on communication standards, such as WiFi, 2G, or 3G, or combinations thereof. In one exemplary embodiment, the wireless communication module 9 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, the wireless communication module 9 further includes a near-field communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0121] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or technical improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A circuit for processing induced voltage differences, characterized in that, include: Excitation current circuit (3); the output terminal of the excitation current circuit (3) is connected to the electrode array (1) which is sleeved on the outside of the lower limb (1-2) of the leg by an elastic fabric band (1-1) through a multiplexing selection circuit (2). The electrode array (1) is also connected to the amplification and calculation circuit (4) through the multiplexing selection circuit (2). The first electrode in the electrode array (1) is located on the outside of the tibialis anterior muscle, and the second electrode is located on the opposite side of the first electrode. The first group of electrodes and the second group of electrodes remaining in the electrode array (1) except for the first electrode and the second electrode are evenly distributed on the outside of the first lower limb arc and the second lower limb arc formed by the first electrode and the second electrode along the outside of the lower limb (1-2). The excitation current circuit (3) is used to provide current excitation to the electrode array (1); the electrode array (1) is used to collect the induced voltage difference corresponding to the internal tissue impedance of the lower leg (1-2); the amplification and calculation circuit (4) is used to amplify the induced voltage difference and calculate the root mean square corresponding to the amplified induced voltage difference. The amplification and calculation circuit (4) includes: a programmable differential instrumentation amplifier circuit (41) connected to the electrode array (1) via the multiplexing selection circuit (2) and a root mean square detector circuit (43) connected to the programmable differential instrumentation amplifier circuit (41). The programmable differential instrument amplifier circuit (41) is used to amplify the induced voltage difference; the root mean square detector circuit is used to calculate the root mean square of the induced voltage difference amplified by the programmable differential instrument amplifier circuit (41).

2. The induced voltage difference processing circuit according to claim 1, characterized in that, The excitation current circuit (3) includes: a signal generator (31) connected to the electrode array (1) via the multiplexing selection circuit (2) and a voltage-to-current conversion circuit (33) connected to the signal generator (31). The signal generator (31) is used to generate an excitation voltage supplied to the electrode array (1); the voltage-to-current conversion circuit (33) is used to convert the excitation voltage into a corresponding current excitation.

3. The induced voltage difference processing circuit according to claim 2, characterized in that, A first filter circuit (32) is configured between the signal generator (31) and the voltage-current conversion circuit (33); wherein the first filter circuit (32) is used to filter the excitation voltage.

4. The induced voltage difference processing circuit according to any one of claims 1-3, characterized in that, A second filter circuit (42) is configured between the programmable differential instrument amplifier circuit (41) and the root mean square detector circuit (43); wherein, the second filter circuit (42) is used to filter the amplified induced voltage difference.

Images